A relative approach for uncertainty quantification of inversely identified material behavior using a heterogeneous test for sheet metal

Abstract

Abstract The process of calibrating the material parameters of a constitutive model has inherent uncertainty sources. These can lead to inadequate material behavior characterization and, consequently, errors in the numerical simulations. A robust mechanical test is necessary for accurate material mechanical characterization. In this work, it is relatively quantified the uncertainties of the parameter identification of the Swift hardening law and the Yld2000-2d yield function using Digital Image Correlation (DIC) and a DIC-levelled approach. The levels of uncertainty are calculated with a derivative-based local method, having into consideration the Karush-Kuhn-Tucker (KKT) conditions for optimization problems. The mechanical test used for the parameter calibration was an optimized interior notched specimen that presents several strain and stress states simultaneously. The synthetic images were generated based on the numerical data, and the strain fields were accessed using DIC techniques. It was concluded that the step size is the DIC parameter that most influences the material parameter identification.